Licență 2017

adiftene@info.uaic.ro

Cuprins

• Mod de lucru

• Partea scrisă (documentația)

• Tematici

• Notarea

Mod de lucru (1)

• Termene limită (3 pe semestru):– 20 Iulie – stabilirea domeniului mare (discuții directe sau

pe e-mail)– 1 Octombrie – documentația – colectați toate elementele

necesare pentru state-of-the-art pe domeniul mare(articole pdf și pagini web pentru descrierea domeniului, definiții importante, aplicații existente (avantaje/dezavantaje), servicii disponibile, resurse disponibile, etc.) => un folder cu toate acestea

– (1) (30) Noiembrie – documentația – primele 20 de pagini– 20 Decembrie – prototip varianta alpha-alpha (servicii,

resurse, module mai importante, etc.)

Mod de lucru (2)

• Termene limită (minim 3 întâlniri pe semestru):– 30 Martie – varianta beta-beta a aplicației

– 30 Mai – varianta beta a documentației (arhitectura aplicației, modulele mai importante, studii de caz, evaluare, comparație cu alte sisteme, etc.)

– 30 Iunie – variantele aproape finale (documentație și aplicație)

• Observații:– Termenele limită se pot negocia (comunicare)

– Nerespectarea lor duce la întreruperea colaborării

Documentația

• Reguli de scriere: http://profs.info.uaic.ro/~adiftene/Scoala/2010/Licenta/ScriereLicenta.txt

• Ghid licență: http://profs.info.uaic.ro/~mdiac/other/licenta2010/ghid_licenta2010.pdf

• Exemple din anii trecuți: http://profs.info.uaic.ro/~adiftene/Licenta/

• Foarte important: drepturile de autor, citările, furtul intelectual

• Când aveți 5-10 pagini scrise le trimiteți pe e-mail pentru a primi feedback și a putea merge mai departe

Tematici (1)

1. Internet of things

a) Case inteligente – controlul de la distanță a AC, centralei termice, jaluzele, etc.

b) Învățarea profilului proprietarului -> acțiuni care să-l ajute (decizii cu o confirmare din partea proprietarului, care poate a uitat să facă anumite setări)

Tematici (2)

2. Augmented Realitya) Jocuri

b) eLearning

c) Ghidare 3D

d) Etc.

Combinat cu GoogleGlasses, Oculus, etc.

http://profs.info.uaic.ro/~adiftene/TeachForFuture/

Tematici (3)

3) Clusterizare la nivel vizual (A1)a) Pentru o colecție dată de imagini se vor extrage

caracteristici vizuale

b) Resurse disponibile: Lire, MathLab, etc.

c) Folosirea de servicii care dau "distanța" dintre douăimagini

d) În funcție de caracteristicile extrase imaginilesimilare se vor grupa pe clustere

e) Librării python pentru clusterizare

Participare la CLEF 2017

Tematici (4)

4) Disruptive technologies (senzori, smart clock, etc.)– Definition: New ways of doing things that disrupt or overturn

the traditional business methods and practices. For example, steam engine in the age of sail, and internet in the age of post office mail.

– Such was the case with Alexander Graham Bell's "electrical speech machine," which we now call the telephone.

– http://whatis.techtarget.com/definition/disruptive-technology

– http://www.mckinsey.com/insights/high_tech_telecoms_internet/disruptive_technologies

– http://www.mckinsey.com/insights/business_technology/disruptive_technologies

– http://en.wikipedia.org/wiki/Disruptive_innovation

Tematici 5 - NLP and Big Data

1) Big-data Idea 1: Linkage of information in a contextual manner.– Go beyond simple publications & grants citation graph.– Infer publication/grant link importance based on the reference position inside the paper (e.g.

abstract, experiment description section, related work, etc.)– Use sentiment analysis for discovering the ‘tone’ of the context.– Example: (1) http://www.math.md/nrofdownloads.php?file=/files/csjm/v23-n2/v23-n2-(pp189-

208).pdf on first page “This work was supported by MUCKE (Multimedia and User Credibility Knowledge Extraction) project Reference No. 2 CHIST-ERA, three years from 01.10.2012.” (2) http://ceur-ws.org/Vol-1391/113-CR.pdf on page 9 “The research presented in this paper was funded by the projectMUCKE (Multimedia and User Credibility Knowledge Extraction), number 2CHIST-ERA/01.10.2012. Special thanks go to all colleagues from the Faculty ofComputer Science, second year, group A1, who were involved in this project.” In case (1) we have a greater importance than in case (2).

• Big-data Idea 2: Finding an advanced similarity metric between publications/grants:– Option 1: Do semantic analysis of the publication texts. Unfortunately, not all data-sources contain

the actual text of the publications.– Option 2: Use advanced graph-theory techniques to infer hidden connections between

publications/grants.

• NLP Idea: Improving publication / grant classification algorithms– Take one of Uber classification algorithms and improve its performance

Tematici 6 - Figshare

• Adaugare automata de publicatii si proiecte pentru UAIC

• Crawling, wrapper,

• https://uaic.figshare.com/

Video Content detection based on close caption , natural language and audio spectogram

• Intention:Design an algorithm that can detect add content in a video stream based on close captions and audio tracks.

Shows aired as live TV streams often are interrupted by inserted ads or commercials.For detection of ads in an automatic way we can use the close caption data. We are proposing an approach to use Natural Language Understanding and machine learning to detect if a piece of text is part of an ads or not.For detection of ads in an automatic way, we can use the sound track. We can use the audio spectrogram and train a SVM or a NN to detect is a segment is part of an add transition

MediaWiki search engine study and proof of concept

• Intention: Improve quality of the search results by increasing the efficiency of the ranking system the search engine uses.MediaWiki is a particular wiki engine that is developed for and used by Wikipedia and the other Wikimedia projects. MediaWiki is freely available for others to use (and improve), and it is in use by all sorts of projects and organizations around the world.One of the most used feature associated is the search engine and the Lucene-search is the one used by large MediaWiki website.

Social Network for Public Transport Monitoring

• Traffic inside the city is becoming more and more hectic, so public transport is an important tool to help the traffic jams. But for public transport to be used at a larger scale, we need to be able to monitor and rely on it. This project proposes to create a social network in which issues, delays and changes in course of trams and busses can be reported by users and displayed to the public, in order to make possible a real time monitoring of the public transports. A nice to have in this project would be an RFID based check-in in public transports of users and correlating the time and traffic flow to measure an approximate time of arrival in different stations.

Alte tematici

• https://docs.google.com/spreadsheets/d/1fIZPeyeddGvV9nFnznoAfFa_9eKMiFRK8fh8zmMN1hs/pubhtml?gid=1429687775&single=true

• http://profs.info.uaic.ro/~adiftene/TeachForFuture/

Tema grea - Orchestrare, coregrafie de servicii. Ontologii

• Folosirea unor servicii atomice situate pe calculatoare diferite

• Aplicatie client-server• Orchestrarea serviciilor din spatele unei diagrame BPMN si

rularea lantului de servicii• Servicii specializate de vreme, localizare, news, retele sociale,

calendar, etc.

• Scenarii:– 1) Citirea din calendar a primului eveniment si aflarea vremii din

ziua respectiva– 2) Citirea unei stiri si localizarea pe GoogleMaps a locatiilor din stire– 3) Localizare + servicii de localizare a restaurantelor din zona

Notarea

• Tematica

• Prezentarea

• Elementele de noutate

• Evaluarea, compararea cu alte soluții

• Coordonatorul

Q&A

Întrebări???